Glossary

Short definitions for terms used across the manual.

Term	Meaning
DataFrameModel	SQLModel-like table class: schema on the class body, generated RowModel, typed transforms
DataFrame[Schema]	Generic typed frame over a Pydantic Schema or compatible model
Schema	Pydantic row shape base type used with DataFrame[T]
Expr	Typed column expression (filters, with_columns, aggregations)
Plan	Lazy logical transform graph executed in Rust/Polars at materialization
ScanFileRoot	Lazy file scan handle from read_* / aread_* — not a materialized table yet
materialize_*	Eager read into dict[str, list] (full file in Python memory)
read_*	Lazy scan root for pipeline transforms before collect or write_*
export_*	Eager write from dict[str, list] to a file
write_* (on DataFrame)	Lazy pipeline sink — executes in Rust without a giant Python dict
collect()	Run the plan; return list of Pydantic row models (default)
to_dict()	Run the plan; return dict[str, list]
trusted_mode	off (full validation), shape_only, or strict (dtype/shape checks)
Materialization mode	Blocking sync, async await, deferred submit, or chunked stream / astream
ExecutionEngine	Protocol for where plans run (default: native Polars-backed extension)
pydantable-native	PyPI package containing pydantable_native._core (Rust extension)
pydantable-protocol	Zero-dependency engine protocols for third-party backends
MissingRustExtensionError	Native extension not built or not loadable
RowModel	Per-row Pydantic model generated from a DataFrameModel (e.g. UserDF.RowModel)

See also: Mental model, I/O decision tree.